Skip to content
Permalink
dd470f2dba
Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?

nmslib/previous_releases_scripts/vldb2015/RunAllL2.pl

Go to file
 
 
Cannot retrieve contributors at this time
executable file 295 lines (250 sloc) 10.8 KB
Raw Blame
#!/usr/bin/perl -w
require "./RunCommon.pm";
my $DataDir=$ENV{"DATA_DIR"};
defined($DataDir) or die("Define the environemnt variable DATA_DIR, which contains data files.");
my $CacheDir=$ENV{"CACHE_DIR"};
defined($CacheDir) or die("Define the environemnt variable CACHE_DIR, which contains gold standard cache files.");
-d $DataDir or die("DATA_DIR='$DataDir' is not a directory");
my $BinDir=$ENV{"BIN_DIR"};
defined($BinDir) or die("Define the environemnt variable BIN_DIR, which contains binary files.");
-d $BinDir or die("BIN_DIR='$BinDir' is not a directory");
-f "$BinDir/experiment" or die("Missing file '$BinDir/experiment'");
my $SpaceType = "L2";
my @LargeRecallList = (0.95, 0.9, 0.85, 0.8, 0.75, 0.7);
my @SmallRecallList = (0.95, 0.90, 0.85);
my @RecallLists = (\@LargeRecallList, \@LargeRecallList,
\@SmallRecallList,
\@SmallRecallList,
\@SmallRecallList,
\@SmallRecallList,
\@SmallRecallList,
\@SmallRecallList);
my @DataSetName = ("cophir", "sift",
"sift",
"sift",
"sift",
"sift",
"sift",
"sift");
my @DataSetFile = ("cophir/full.txt", "sift_texmex_learn5m.txt",
"sift_texmex_learn5m.txt",
"sift_texmex_learn5m.txt",
"sift_texmex_learn5m.txt",
"sift_texmex_learn5m.txt",
"sift_texmex_learn5m.txt",
"sift_texmex_learn5m.txt");
my @CachePrefix = ("cophir5M", "sift5M",
"sift2.5M",
"sift1.25M",
"sift625K",
"sift300K",
"sift150K",
"sift75K");
my @TuneParamFilePrefix=("tunning/ResultsL2/OutFile.cophir", "tunning/ResultsL2/OutFile.sift",
"tunning/ResultsL2/OutFile.sift",
"tunning/ResultsL2/OutFile.sift",
"tunning/ResultsL2/OutFile.sift",
"tunning/ResultsL2/OutFile.sift",
"tunning/ResultsL2/OutFile.sift",
"tunning/ResultsL2/OutFile.sift");
my @MaxNumData = (5000000, 5000000,
2500000,
1250000,
625000,
300000,
150000,
75000);
my @NN = (10, 13,
13,
13,
13,
13,
13,
13);
my @BucketSize = (50, 50,
50,
50,
50,
50,
50,
50);
my @IndexAttempts = (3, 3,
3,
3,
3,
3,
3,
3);
my @SearchAttempts = ([1..5,8,11,15,20,25,30,35,40,45,50], [1..5,8,11,15,20,25,30],
[1..10],
[1..10],
[1..10],
[1..10],
[1..10],
[1..10]);
my @NumPivot =(2000, 2000,
2000,
2000,
2000,
2000,
2000,
2000
);
my @NumPivotIndex =(20, 20,
20,
20,
20,
20,
20,
20,
);
my @NumPivotSearch =([4,5,6,7,8,9,10], [4,5,6,7,8,9,10],
[4,5,6,7,8,9,10],
[4,5,6,7,8,9,10],
[4,5,6,7,8,9,10],
[4,5,6,7,8,9,10],
[4,5,6,7,8,9,10],
[4,5,6,7,8,9,10],
);
my @DbScanFrac =([], [],
[],
[],
[],
[],
[],
[]
);
my @MaxNumQuery = (1000, 500,
500,
500,
500,
500,
500,
500);
my @TestSetQty = (5, 1,
1,
1,
1,
1,
1,
1);
my @Use = (1, 1,
0,
0,
0,
0,
0,
0);
my $DataSetQty = scalar(@DataSetName);
scalar(@RecallLists) == $DataSetQty or die("RecallList has wrong number of elems");
scalar(@NumPivot) == $DataSetQty or die("NumPivot has wrong number of elems");
scalar(@NumPivotIndex) == $DataSetQty or die("NumPivotIndex has wrong number of elems");
scalar(@NumPivotSearch) == $DataSetQty or die("NumPivotSearch has wrong number of elems");
scalar(@DbScanFrac) == $DataSetQty or die("DbScanFrac has wrong number of elems");
scalar(@DataSetFile) == $DataSetQty or die("DataSetFile has wrong number of elems");
scalar(@MaxNumData) == $DataSetQty or die("MaxNumData has wrong number of elems");
scalar(@MaxNumQuery) == $DataSetQty or die("MaxNumQuery has wrong number of elems");
scalar(@CachePrefix) == $DataSetQty or die("CachePrefix has wrong number of elems");
scalar(@Use) == $DataSetQty or die("Use has wrong number of elems");
scalar(@TestSetQty) == $DataSetQty or die("TestSetQty has wrong number of elems");
scalar(@NN) == $DataSetQty or die("NN has wrong number of elems");
scalar(@IndexAttempts)== $DataSetQty or die("IndexAttempts has wrong number of elems");
scalar(@SearchAttempts)== $DataSetQty or die("SearchAttempts has wrong number of elems");
#RunTest(1);
RunTest(10);
sub RunTest {
my ($K) = @_;
for (my $dn = 0; $dn < $DataSetQty; ++$dn) {
my $Name = $DataSetName[$dn];
my $DataQty = $MaxNumData[$dn];
if (!$Use[$dn]) {
print "Skipping $Name K=$K (qty=$DataQty)\n";
next;
}
my $OutFileDir="ResultsL2/$Name";
!system("mkdir -p $OutFileDir") or die("Cannot create $OutFileDir");
my $OutFilePrefix="$OutFileDir/res";
if (1) {
RunNAPP($BinDir, $DataDir, $CacheDir,
$SpaceType,
$DataSetFile[$dn],
$DataQty,
$TestSetQty[$dn],
$MaxNumQuery[$dn],
$CachePrefix[$dn],
$K, $OutFilePrefix,
$NumPivot[$dn],
$NumPivotIndex[$dn],
$NumPivotSearch[$dn],
$DbScanFrac[$dn]
);
}
if (1) {
RunVPTree($BinDir, $DataDir, $CacheDir,
$SpaceType,
$DataSetFile[$dn],
$DataQty,
$TestSetQty[$dn],
$MaxNumQuery[$dn],
$CachePrefix[$dn],
$K, $OutFilePrefix,
$TuneParamFilePrefix[$dn],
$BucketSize[$dn]);
}
if (1) {
RunSmallWorld($BinDir, $DataDir, $CacheDir,
$SpaceType,
$DataSetFile[$dn],
$DataQty,
$TestSetQty[$dn],
$MaxNumQuery[$dn],
$CachePrefix[$dn],
$K, $OutFilePrefix,
$NN[$dn],
$IndexAttempts[$dn],
$SearchAttempts[$dn]);
}
if (1) {
RunMPLSH($BinDir, $DataDir, $CacheDir,
$SpaceType,
$DataSetFile[$dn],
$DataQty,
$TestSetQty[$dn],
$MaxNumQuery[$dn],
$CachePrefix[$dn],
$K, $OutFilePrefix,
$RecallLists[$dn]);
}
}
}
sub RunMPLSH {
my ($BinDir, $DataDir, $CacheDir,
$SpaceType,
$DataNameLoc,
$MaxNumDataLoc, $TestSetQtyLoc, $MaxNumQueryLoc,
$CachePrefixLoc,
$KLoc,
$OutFilePrefixLoc,
$RecallListLocRef) = @_;
defined($BinDir) or die("undefined BinDir");
defined($BinDir) or die("undefined BinDir");
defined($CacheDir) or die("undefined CacheDir");
defined($RecallListLocRef)or die("undefined RecallListLocRef");
# These two values provide nearly optimal performance on all our data sets.
my $L = 50;
my $T = 10;
for my $recall (@{$RecallListLocRef}) {
my $H = $MaxNumDataLoc + 1; # Works really well and is memory efficient
my @MethArray = ();
push(@MethArray, " --method lsh_multiprobe:desiredRecall=$recall,H=$H,T=$T,L=$L,tuneK=$KLoc");
# Run each one in a separate binary, or else we'll run out of memory!
RunCmd($BinDir, $DataDir, $CacheDir,
$SpaceType,
$DataNameLoc, $MaxNumDataLoc, $TestSetQtyLoc,
$MaxNumQueryLoc,
$CachePrefixLoc,
$KLoc,
$OutFilePrefixLoc,
\@MethArray);
}
}